Decision Trees - Machine Learning
Essay by Angad Jabbal • April 7, 2017 • Coursework • 2,211 Words (9 Pages) • 1,140 Views
Marketing Model for Brand Advocacy
Customer Loyalty Assessment based on Demographics
[pic 1]
Model Summary | Decision Trees – CHAID Approach | |||||||
Response & Predictors | ||||||||
Specifications | Growing Method | CHAID | Is the tree model a good predictive model for High Loyalty Customers? | |||||
Name | Type | Values | ||||||
Dependent Variable | BINARY LOYALTY | |||||||
Loyalty | Ordinal | 0-LOW | ||||||
Independent Variables | AGE, GENDER, | 1-HIGH | ||||||
EDUCATION, MARITAL | 1:18-24 | |||||||
STATUS, INCOME | 2:25-34 | |||||||
Age | Ordinal | 3:35-44 | ||||||
4:45-54 | ||||||||
5:55+years | ||||||||
Validation | None | 1-MALE | ||||||
Gender | Ordinal | |||||||
Maximum Tree Depth | 3 | 2-FEMALE | ||||||
Minimum Cases in | 25 | 1-High School | ||||||
2- Some College | ||||||||
Education | Ordinal | |||||||
Parent Node | 3-College | |||||||
4-Graduate School | ||||||||
Minimum Cases in | 15 | 1-Married | ||||||
Marital | ||||||||
Child Node | Ordinal | 2-Single | ||||||
Status | 3-Divorced/Separated | 4- | ||||||
Results | Independent Variables | GENDER | Widow/Widower | |||||
Included | 1:<35K | |||||||
Number of Nodes | 3 | 2:35-54.999K | ||||||
3:55-74.999K | ||||||||
Number of Terminal | 2 | Income | Ordinal | 4:75-94.999K | ||||
5-95-114.999K | ||||||||
Nodes | 6:115-134.999K | |||||||
7:>135K | ||||||||
Depth | 1 | |||||||
Customer Loyalty Assessment based on Demographics
[pic 2]
- Among predictors i.e. the demographic variables - Assessing based on the CHAID Analysis yields that only Gender is the predictor that gives a Chi Square value having statistical significance < .05 or 5% for the relation of its 2 categories with response variable 'Loyalty'. This means that the process of comparison between 2 categories of a predictor, repeated for each of the predictors merging the categories found not to have statistically significant relationship with response variable in earlier comparisons did not have any 2 categories of a predictor having a significant effect on loyalty of the customer for the brand except for Male and Female categories of Gender that show significant affect on customer loyalty.
- As per the CHAID algorithm, the categories for Ordinal Variables are naturally defined and for continuous variables are taken in a bucket of 10 units(such as for Age).
[pic 3][pic 4]
Customer Loyalty Assessment based on Demographics
[pic 5][pic 6][pic 7][pic 8][pic 9][pic 10][pic 11][pic 12][pic 13][pic 14][pic 15][pic 16][pic 17][pic 18][pic 19][pic 20][pic 21][pic 22][pic 23][pic 24]
...
...